Open Source Speech Recognition Engines

NeMo is a multimodal AI framework and toolkit designed for the development, training, and scaling of large language models, generative AI systems, and speech-based models. It functions as an automatic speech recognition toolkit, a text-to-speech engine, and a framework for building models that process and generate combinations of text, image, and audio data. The project serves as a conversational AI orchestrator capable of managing real-time, interruptible voice interactions. It provides specialized workflows for speech translation, converting spoken audio from one language into text or speech in another. The platform covers a broad range of AI model development capabilities, including the training of generative and speech models. Its operational surface includes automatic speech recognition, text-to-speech synthesis, and the creation of multimodal pipelines.

NeMo is a comprehensive framework designed for the development, training, and deployment of large-scale conversational and generative artificial intelligence models. It provides an integrated platform for building multimodal systems, encompassing speech processing, language modeling, and reinforcement learning alignment. The framework is built to handle the entire lifecycle of AI development, from data curation and model pretraining to production-ready service deployment. The platform distinguishes itself through advanced distributed training capabilities, including tensor and pipeline parallelism, which allow for the execution of models that exceed the memory capacity of individual hardware devices. It incorporates specialized architectures such as mixture-of-experts to optimize computational efficiency and includes a programmable guardrails system to enforce safety policies and topical boundaries on model outputs. Additionally, the framework supports retrieval-augmented generation to ground model responses in external knowledge bases, reducing hallucinations and improving factual accuracy. Beyond core training and inference, the framework offers extensive tools for audio signal processing, speech-to-text transcription, and text-to-speech

This project is a multimodal translation framework and large language model capable of speech-to-speech, speech-to-text, and text-to-text translation across nearly 100 languages. It provides a real-time speech translation engine and a comprehensive toolkit for converting spoken audio between languages. The system is distinguished by its ability to preserve the original speaker's tone, pace, and prosody during translation. It utilizes a specialized on-device inference toolkit that converts model checkpoints into C-based libraries, enabling low-latency execution on mobile and edge hardware without a Python runtime. The framework covers a wide range of capabilities including automatic speech recognition, expressive speech synthesis, and real-time translation streaming. It also includes audio content moderation for toxicity detection and tools for multimodal translation evaluation and distributed model fine-tuning. The project is implemented using Jupyter Notebooks.

Omnilingual-ASR is a multilingual automatic speech recognition framework and toolkit designed to transcribe audio across 1,600 languages. It provides a complete pipeline for converting speech to text, including a toolkit for fine-tuning pre-trained speech models to specific languages or datasets using custom training recipes. The system supports zero-shot speech recognition, allowing the model to predict text in unseen languages without extensive training data. It further enables few-shot language guidance through in-context examples and uses language codes to constrain transcription output to the correct target language and script. The framework includes capabilities for high-throughput transcription via parallelized batch processing and a modular audio pipeline that normalizes and resamples diverse input formats. Resource management is handled through a system of asset cards and a command-line interface for retrieving metadata related to models, datasets, and tokenizers.

faster-whisper is an automatic speech recognition framework and an optimized implementation of the Whisper speech-to-text engine. It functions as a CTranslate2 inference engine designed to convert spoken audio into written text. The project serves as a model quantization tool that transforms large audio model weights into lower precision formats. This process reduces memory usage and increases execution speed on hardware by utilizing integer quantized weights. The framework covers a broad range of capabilities including batch audio transcription for parallel processing and voice activity detection to filter out non-speech audio segments. It also provides utilities for converting original or fine-tuned audio models into formats compatible with the CTranslate2 runtime.

ESPnet is a comprehensive speech processing toolkit and PyTorch-based trainer designed for building end-to-end speech recognition, synthesis, and translation models. It provides a structured framework for developing automatic speech recognition systems using transducer and encoder-decoder architectures, alongside engines for text-to-speech synthesis and speech translation pipelines. The project distinguishes itself through a recipe-based workflow execution system that ensures experimental reproducibility by running standardized sequences of scripts for data preparation and model training. It leverages containerized environments to provide consistent execution across platforms and supports large-scale distributed training across multiple GPUs and nodes. The toolkit covers a broad range of capabilities, including spoken language understanding for intent and sentiment classification, audio enhancement and separation, and singing voice synthesis. It also incorporates advanced training techniques such as self-supervised learning, parameter-efficient fine-tuning, and transfer learning. Model development is supported by utilities for audio data formatting, spectral augmentation, and the integration of pretrained encoders, while inference is optimized through blockwise beam search for real-time streaming execution.

Ten Framework is a multimodal large language model agent framework designed for building low-latency conversational agents. It integrates voice, text, and visual inputs in real time to facilitate human interaction. The project includes a real-time speech processing pipeline for streaming transcription, voice activity detection, and speaker diarization. It also features an avatar synchronization engine that coordinates character lip animations and visual outputs with synthesized speech. The framework covers edge AI deployment through containerized packaging and direct integration with embedded hardware boards. Additional capabilities include a telephony gateway for connecting agents to phone networks via the Session Initiation Protocol and tools for real-time visual generation of sketches and doodles.

Pydub is a Python audio manipulation library and digital audio processor used for editing, slicing, and converting audio files and segments. It serves as a programmatic wrapper for FFmpeg to import and export a wide variety of audio formats. The library functions as an audio signal generator capable of creating synthetic waveforms, such as sine waves and white noise. It also provides tools for digital signal processing, including the application of filters, fades, crossfades, and gain adjustments to sound signals. Its broader capabilities cover programmatic audio editing through concatenation and mixing, automated audio analysis for metadata extraction and silence detection, and sample format conversion. The toolkit also supports frequency filtering, DC offset removal, and the generation of silent segments for spacing. The library can route processed audio segments to system speakers using external playback drivers.

This Python SDK provides a comprehensive toolkit for synthetic audio generation, voice cloning, and the development of conversational AI agents. It enables the creation of lifelike spoken audio from text, the replication of human voices through custom cloning, and the deployment of real-time voice agents capable of interacting with external large language models. The library distinguishes itself through deep integration of conversational AI capabilities, including the design of agent personas and the execution of real-time actions via APIs. It supports professional-grade audio production through a variety of specialized tools for multilingual dubbing, studio-quality music generation, and high-fidelity sound effects. The SDK covers a broad surface of speech and media processing, including real-time audio streaming via WebSockets, speech-to-text transcription with speaker diarization, and the synchronization of audio with visual elements. It also provides utilities for monitoring generation costs and managing agent security through response guardrails and access controls.

Vocode-core is a framework for building real-time conversational AI voice agents. It serves as a conversational orchestrator and pipeline that integrates speech-to-text, large language models, and text-to-speech services to enable low-latency voice interactions. The project features a provider-agnostic interface that allows for swappable speech and language model providers, including support for both cloud APIs and local binaries. It distinguishes itself through a specialized telephony integration layer that enables agents to be deployed across phone lines, WebRTC, and virtual meeting platforms. The framework covers a broad range of capabilities including agent orchestration with custom personas and tool assignments, real-time audio streaming with interruption handling, and comprehensive telephony management for inbound and outbound call lifecycles. It also includes speech processing tools for multi-language transcription, synthetic voice cloning, and event-driven webhooks for monitoring call milestones.

This project is an on-device AI SDK providing a framework for running large language models, vision models, and speech models locally. It serves as an orchestration layer for local LLM execution, ensuring data privacy and offline availability by utilizing hardware acceleration on the device. The SDK is distinguished by its comprehensive voice and multimodal capabilities, including a coordinated voice pipeline for activity detection, speech-to-text, and text-to-speech synthesis. It also provides a dedicated implementation kit for local retrieval-augmented generation and tools for processing combined image and text inputs via vision-language models. The broader capability surface covers model lifecycle management, including downloading, caching, and the dynamic swapping of fine-tuned adapters. It includes support for structured output generation, tool calling for external function integration, and hardware-accelerated image generation. The system also incorporates performance monitoring for inference metrics and comprehensive audio-visual capture tools for camera and microphone input.

GPT4All is a cross-platform runtime environment designed to execute large language models directly on local consumer hardware. By leveraging an optimized C++ inference backend, it enables private, offline AI interactions without requiring an internet connection or external cloud services. The project provides a comprehensive ecosystem for managing the entire model lifecycle, including discovery, downloading, and configuration of local weights. What distinguishes the platform is its integrated retrieval-augmented generation engine, which allows users to index local documents into semantic vector spaces. This capability enables context-aware chat sessions where the model can reference private files, notes, and spreadsheets to provide grounded, relevant responses. The system also features a local HTTP server that exposes an OpenAI-compatible API, allowing developers to integrate these private, self-hosted models into existing applications and workflows. Beyond its core inference and retrieval capabilities, the project includes a graphical desktop interface for end-user interaction and a Python software development kit for programmatic access. These tools support advanced configuration of model parameters, performance monitoring, and the management of local embedding pipelines for custom semantic search tasks. The software is distributed as a unified application package, with documentation available to guide users through installation and local environment setup.

LiveKit is a comprehensive framework for building and orchestrating real-time, multimodal AI agents that interact with users through voice, video, and text. It provides a centralized, event-driven architecture to manage the entire lifecycle of automated participants, from initialization and session state management to graceful shutdown. By utilizing a selective forwarding unit, the platform efficiently routes media streams between participants and agents, ensuring low-latency communication and secure, token-based authentication for all connections. The platform distinguishes itself through its modular pipeline-based media processing, which chains specialized speech-to-text, language, and text-to-speech services into cohesive workflows. It includes advanced capabilities for real-time voice activity detection, enabling natural turn-taking and interruption handling, alongside remote procedure call tooling that allows agents to execute external functions or access local resources during a conversation. Developers can further extend these interactions by integrating photorealistic virtual avatars that synchronize visual expressions with the agent's audio output. Beyond core conversational logic, the system offers extensive support for telephony integration, allowing agents to connect to public networks via SIP for inbound and outbound calling. It provides a robust suite of observability and monitoring tools to track agent performance, connection quality, and session events, ensuring reliability in production environments. The platform also includes specialized utilities for task automation, such as capturing and validating structured user data, and supports multi-step workflow orchestration to handle complex, context-aware interactions. The project provides a command-line interface for scaffolding, deploying, and testing agent applications, with documentation available in machine-readable formats to assist in development.

Pipecat is a framework and software development kit for building real-time multimodal AI agents and speech-to-speech systems. It utilizes a frame-based data pipeline to route audio, video, and text through a modular sequence of processors, enabling the orchestration of low-latency conversational AI. The project is distinguished by its ability to coordinate complex multimodal services, including speech-to-text, language models, and text-to-speech, within a single pipeline. It features semantic voice activity detection for natural turn-taking, state-machine conversation flows for dialogue management, and WebRTC-based streaming for bidirectional media connectivity. The framework covers a broad surface of capabilities, including AI integration with various foundation models, asynchronous tool execution for external function calls, and telephony integration with providers such as Twilio and Genesys Cloud. It also includes tools for distributed session management, long-term agent memory, and cloud deployment orchestration for scaling agent instances. The project provides command-line utilities for project scaffolding, deployment auditing, and technical documentation indexing.